import json
import csv

# For processing json watch history into a dataset

JSON_FILE = "watch-history.json"
CSV_FILE = "watch-history.csv"

# load in file
with open(JSON_FILE, 'r', encoding='utf-8') as file:
    data = json.load(file)

# Extract columns data
with open(CSV_FILE, mode='w', newline='', encoding='utf-8') as csv_file:
    # open output file
    fout = csv.writer(csv_file)
    fout.writerow(['Title', 'Video URL', 'Channel', 'Date Watched', 'Ad?'])

    # process each entry
    for item in data:
        title = item.get('title', 'No Title')
        video_url = item.get('titleUrl', 'No URL')
        date = item.get('time', 'No Date')
        ad_status = False

        # Try to get channel, json includes youtube ads so those either
        # have a channel or are N/A
        if 'subtitles' in item:
            channel_name = item['subtitles'][0].get('name', 'N/A')
        else:
            channel_name = 'N/A'

        # further checks for ad status
        if 'details' in item:
            ad_status = True

        # write data row
        fout.writerow([title, video_url, channel_name, date, ad_status])

import pandas as pd
from datetime import datetime, timedelta
import matplotlib.pyplot as plt
import seaborn as sns
import re
import scipy.stats as stats

#import the first data set of the channel's watch history
df = pd.read_csv("watch-history.csv")
df.head(10)

filtered_df = df[df["Ad?"] != True]
filtered_df = filtered_df[~filtered_df["Title"].str.startswith("Viewed ")].reset_index(drop=True)
filtered_df["Title"] = filtered_df["Title"].str.replace("Watched ", "", n=1)

filtered_df

filtered_df.to_csv("filtered_history.csv")

import json
import csv
import re

from googleapiclient.discovery import build
from config import API_KEY

youtube = build('youtube', 'v3', developerKey=API_KEY)

#makes the api call and fetches data categories
def fetch(video):
    request = youtube.videos().list(part="snippet,statistics,contentDetails,topicDetails", id=video)
    response = request.execute()

    if response:
        if response["items"]: #deals with privated or videos that no longer exist TwosetViolin RIP :(
            v = response["items"][0]

            data = {
                "title": v["snippet"]["title"],
                "description": v["snippet"]["description"],
                "channel_name": v["snippet"]["channelTitle"],
                "published_at": v["snippet"]["publishedAt"],
                "category_id": v["snippet"]["categoryId"],
                "view_count": v["statistics"].get("viewCount", "N/A"),
                "like_count": v["statistics"].get("likeCount", "N/A"),
                "comment_count": v["statistics"].get("commentCount", "N/A"),
                "duration": v["contentDetails"]["duration"]
            }

            return data
    return None

#uses category id and decodes into actual category
def fetch_category(category_id):
    request = youtube.videoCategories().list(part="snippet", id=category_id)
    response = request.execute()

    if response["items"]:
        category = response["items"][0]
        return category["snippet"]["title"]
    return "N/A"

#takes in a given csv and fetches additional info to place into output csv
def process(input_csv, output_csv):
    with open(input_csv, 'r', encoding='utf-8') as infile, open(output_csv, 'w', newline='', encoding='utf-8') as outfile:
        reader = csv.DictReader(infile)
        writer = csv.DictWriter(outfile, fieldnames=["Title", "Video URL", "Channel", "Category", "View Count", "Like Count", "Comment Count", "Duration"])
        writer.writeheader()

        for row in reader:
            url = row["Video URL"]
            match = re.search(r"v=([^&]+)", url)
            if match: #potentially if no url is included in the file
                video_id = match.group(1)

                if video_id:
                    video_details = fetch(video_id)

                    #ensuring that fetch had a result
                    if video_details:
                        category_name = fetch_category(video_details["category_id"])

                        writer.writerow({
                            "Title": video_details["title"],
                            "Video URL": url,
                            "Channel": video_details["channel_name"],
                            "Category": category_name,
                            "View Count": video_details["view_count"],
                            "Like Count": video_details["like_count"],
                            "Comment Count": video_details["comment_count"],
                            "Duration": video_details["duration"]
                        })

process('filtered_history.csv', 'videos6.csv')

videos = pd.read_csv("videos_dataset.csv")
videos

len1 = filtered_df.shape[0]
len2 = videos.shape[0]  #gets number of rows in each dataframe

print(f'Length of filtered_df: {len1}')
print(f'Length of videos: {len2}')

print(f'Difference: {len1-len2}') # the difference in number

Length of filtered_df: 30838
Length of videos: 28419
Difference: 2419

no_url = filtered_df[filtered_df["Video URL"]=="No URL"]
no_url

filtered_df = filtered_df[filtered_df["Video URL"] != "No URL"]
not_present = filtered_df[~filtered_df["Video URL"].isin(videos["Video URL"])]
not_present

filtered_df = filtered_df[filtered_df["Video URL"].isin(videos["Video URL"])]
watch_history = filtered_df.reset_index(drop=True) #renaming just for clarity

videos = videos.join(watch_history["Date Watched"])
videos["Date Watched"] = pd.to_datetime(videos["Date Watched"], format='ISO8601')
videos

# Possible time examples: PT34S, PT1H13M2S, PT42M51S
def to_minutes(duration):
    total = 0

    hours = re.search(r'(\d+)H', duration)
    minutes = re.search(r'(\d+)M', duration)
    seconds = re.search(r'(\d+)S', duration)

    # If matches are found add to sum
    if hours:
        total += 60 * int(hours.group(1))
    if minutes:
        total += int(minutes.group(1))
    if seconds:
        total += int(seconds.group(1)) / 60

    return total

videos["Duration (Min)"] = videos["Duration"].apply(to_minutes)
videos

videos.info()
videos.describe()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 28419 entries, 0 to 28418
Data columns (total 10 columns):
 #   Column          Non-Null Count  Dtype              
---  ------          --------------  -----              
 0   Title           28419 non-null  object             
 1   Video URL       28419 non-null  object             
 2   Channel         28419 non-null  object             
 3   Category        28419 non-null  object             
 4   View Count      28393 non-null  float64            
 5   Like Count      28373 non-null  float64            
 6   Comment Count   27778 non-null  float64            
 7   Duration        28419 non-null  object             
 8   Date Watched    28419 non-null  datetime64[ns, UTC]
 9   Duration (Min)  28419 non-null  float64            
dtypes: datetime64[ns, UTC](1), float64(4), object(5)
memory usage: 2.2+ MB

cat_counts = videos["Category"].value_counts()

cat_counts.plot(kind="bar")
plt.title("Counts of Category Types")
plt.ylabel("Count")
plt.xlabel("Category")

Text(0.5, 0, 'Category')

grouped = videos.groupby("Category")[["View Count", "Like Count", "Comment Count"]].mean()
grouped

view_anova = stats.f_oneway(videos[videos["Category"]=="Autos & Vehicles"]["View Count"], videos[videos["Category"]=="Comedy"]["View Count"], videos[videos["Category"]=="Education"]["View Count"], videos[videos["Category"]=="Entertainment"]["View Count"], videos[videos["Category"]=="Film & Animation"]["View Count"], videos[videos["Category"]=="Gaming"]["View Count"], videos[videos["Category"]=="Howto & Style"]["View Count"], videos[videos["Category"]=="Music"]["View Count"], videos[videos["Category"]=="News & Politics"]["View Count"], videos[videos["Category"]=="Nonprofits & Activism"]["View Count"], videos[videos["Category"]=="People & Blogs"]["View Count"], videos[videos["Category"]=="Pets & Animals"]["View Count"], videos[videos["Category"]=="Science & Technology"]["View Count"], videos[videos["Category"]=="Sports"]["View Count"], videos[videos["Category"]=="Travel & Events"]["View Count"], nan_policy="omit")
like_anova = stats.f_oneway(videos[videos["Category"]=="Autos & Vehicles"]["Like Count"], videos[videos["Category"]=="Comedy"]["Like Count"], videos[videos["Category"]=="Education"]["Like Count"], videos[videos["Category"]=="Entertainment"]["Like Count"], videos[videos["Category"]=="Film & Animation"]["Like Count"], videos[videos["Category"]=="Gaming"]["Like Count"], videos[videos["Category"]=="Howto & Style"]["Like Count"], videos[videos["Category"]=="Music"]["Like Count"], videos[videos["Category"]=="News & Politics"]["Like Count"], videos[videos["Category"]=="Nonprofits & Activism"]["Like Count"], videos[videos["Category"]=="People & Blogs"]["Like Count"], videos[videos["Category"]=="Pets & Animals"]["Like Count"], videos[videos["Category"]=="Science & Technology"]["Like Count"], videos[videos["Category"]=="Sports"]["Like Count"], videos[videos["Category"]=="Travel & Events"]["Like Count"], nan_policy="omit")
comment_anova = stats.f_oneway(videos[videos["Category"]=="Autos & Vehicles"]["Comment Count"], videos[videos["Category"]=="Comedy"]["Comment Count"], videos[videos["Category"]=="Education"]["Comment Count"], videos[videos["Category"]=="Entertainment"]["Comment Count"], videos[videos["Category"]=="Film & Animation"]["Comment Count"], videos[videos["Category"]=="Gaming"]["Comment Count"], videos[videos["Category"]=="Howto & Style"]["Comment Count"], videos[videos["Category"]=="Music"]["Comment Count"], videos[videos["Category"]=="News & Politics"]["Comment Count"], videos[videos["Category"]=="Nonprofits & Activism"]["Comment Count"], videos[videos["Category"]=="People & Blogs"]["Comment Count"], videos[videos["Category"]=="Pets & Animals"]["Comment Count"], videos[videos["Category"]=="Science & Technology"]["Comment Count"], videos[videos["Category"]=="Sports"]["Comment Count"], videos[videos["Category"]=="Travel & Events"]["Comment Count"], nan_policy="omit")

print("ANOVA for View Count:")
print("F-statistic:", view_anova.statistic)
print("p-value:", view_anova.pvalue)

print("\nANOVA for Like Count:")
print("F-statistic:", like_anova.statistic)
print("p-value:", like_anova.pvalue)

print("\nANOVA for Comment Count:")
print("F-statistic:", comment_anova.statistic)
print("p-value:", comment_anova.pvalue)

ANOVA for View Count:
F-statistic: 60.500591309938855
p-value: 3.982746306618048e-169

ANOVA for Like Count:
F-statistic: 51.617927068847294
p-value: 2.999237540094993e-143

ANOVA for Comment Count:
F-statistic: 34.1540360427735
p-value: 2.720894597885004e-92

plt.figure(figsize=(8, 8))
plt.ylim()
plt.yscale('log')
plt.boxplot([videos[videos["Category"] == cat]["View Count"][videos["View Count"] > 0] for cat in videos["Category"].unique()], labels=videos["Category"].unique())
plt.ylabel("Count")
plt.xlabel("Category")
plt.title("Comparing Views across Category")
plt.xticks(rotation=90)
plt.show()

plt.figure(figsize=(8, 8))
plt.ylim()
plt.yscale('log')
plt.boxplot([videos[videos["Category"] == cat]["Like Count"][videos["Like Count"] > 0] for cat in videos["Category"].unique()], labels=videos["Category"].unique())
plt.ylabel("Count")
plt.xlabel("Category")
plt.title("Comparing Likes across Category")
plt.xticks(rotation=90)
plt.show()

plt.figure(figsize=(8, 8))
plt.ylim()
plt.yscale('log')
plt.boxplot([videos[videos["Category"] == cat]["Comment Count"][videos["Comment Count"] > 0] for cat in videos["Category"].unique()], labels=videos["Category"].unique())
plt.ylabel("Count")
plt.xlabel("Category")
plt.title("Comparing Comments across Category")
plt.xticks(rotation=90)
plt.show()

corr_mat = videos[["View Count", "Like Count", "Comment Count"]].corr()
sns.heatmap(corr_mat, annot=True, cmap="mako")
plt.title("Correlation Heatmap for View, Like, and Comment Count")
plt.show()

recent_videos = videos[videos["Date Watched"] >= videos["Date Watched"].max() - pd.DateOffset(years=1)].reset_index(drop=True)

old_videos = videos[videos["Date Watched"] <= videos["Date Watched"].min() + pd.DateOffset(years=1)].reset_index(drop=True)

res = stats.ttest_ind(old_videos["Duration (Min)"], recent_videos["Duration (Min)"])

print("T test statistic: ", res.statistic)
print("p value: ", res.pvalue)

T test statistic:  -20.976805862045513
p value:  8.398484775659094e-96

recent_avgs = recent_videos.groupby([recent_videos["Date Watched"].dt.year, recent_videos["Date Watched"].dt.month])["Duration (Min)"].mean().reset_index(drop=True)
old_avgs = old_videos.groupby([old_videos["Date Watched"].dt.year, old_videos["Date Watched"].dt.month])["Duration (Min)"].mean().reset_index(drop=True)

plt.plot(recent_avgs, label="Recent Videos", marker="o")
plt.plot(old_avgs, label='Old Videos', marker="x")
plt.title("Average Video Duration Over a Year (Old vs Recent)")
plt.xlabel("Month in the Year")
plt.ylabel("Average Duration (Min)")
plt.legend(title="Video Group")
plt.show()

from sklearn.cluster import KMeans
from sklearn.preprocessing import StandardScaler
import numpy as np

features = videos[["View Count", "Like Count", "Comment Count", "Duration (Min)"]]
categories_encoded = pd.get_dummies(videos['Category'])
features = pd.concat([features, categories_encoded], axis=1)
features

features["hour_sin"] = np.sin(2 * np.pi * videos["Date Watched"].dt.hour / 24)
features["hour_cos"] = np.cos(2 * np.pi * videos["Date Watched"].dt.hour / 24)
features["day_sin"] = np.sin(2 * np.pi * videos["Date Watched"].dt.dayofweek / 7)
features["day_cos"] = np.cos(2 * np.pi * videos["Date Watched"].dt.dayofweek / 7)
features["month__sin"] = np.sin(2 * np.pi * videos["Date Watched"].dt.month / 12)
features["month_cos"] = np.cos(2 * np.pi * videos["Date Watched"].dt.month / 12)
features["year"] = videos["Date Watched"].dt.year
features

print(features.isna().sum())

View Count                26
Like Count                46
Comment Count            641
Duration (Min)             0
Autos & Vehicles           0
Comedy                     0
Education                  0
Entertainment              0
Film & Animation           0
Gaming                     0
Howto & Style              0
Music                      0
News & Politics            0
Nonprofits & Activism      0
People & Blogs             0
Pets & Animals             0
Science & Technology       0
Sports                     0
Trailers                   0
Travel & Events            0
hour_sin                   0
hour_cos                   0
day_sin                    0
day_cos                    0
month__sin                 0
month_cos                  0
year                       0
dtype: int64

features["View Count"] = features["View Count"].fillna(features["View Count"].median())
features["Like Count"] = features["Like Count"].fillna(features["Like Count"].median())
features["Comment Count"] = features["Comment Count"].fillna(features["Comment Count"].median())
print(features.isna().sum())

View Count               0
Like Count               0
Comment Count            0
Duration (Min)           0
Autos & Vehicles         0
Comedy                   0
Education                0
Entertainment            0
Film & Animation         0
Gaming                   0
Howto & Style            0
Music                    0
News & Politics          0
Nonprofits & Activism    0
People & Blogs           0
Pets & Animals           0
Science & Technology     0
Sports                   0
Trailers                 0
Travel & Events          0
hour_sin                 0
hour_cos                 0
day_sin                  0
day_cos                  0
month__sin               0
month_cos                0
year                     0
dtype: int64

scaler = StandardScaler()
features_scaled = scaler.fit_transform(features)

kmeans = KMeans(n_clusters=3, random_state=42)
kmeans.fit(features_scaled)
features["Cluster"] = kmeans.labels_

from sklearn.decomposition import PCA
import colorcet  as cc
palette = sns.color_palette(cc.glasbey, n_colors=3)

pca = PCA(n_components=2)
features_pca = pca.fit_transform(features_scaled)

sns.scatterplot(x=features_pca[:, 0],y=features_pca[:, 1], hue="Cluster", data=features,s=45,legend="full", palette=palette)
plt.title("Cluster Distribution")
plt.xlabel("PCA1")
plt.ylabel("PCA2")

Text(0, 0.5, 'PCA2')

from sklearn.manifold import TSNE

tsne = TSNE(n_components=2, random_state=42)
features_tsne = tsne.fit_transform(features_scaled)

sns.scatterplot(x=features_tsne[:, 0],y=features_tsne[:, 1], hue="Cluster", data=features,s=45,legend="full", palette=palette)
plt.title("Cluster Distribution")
plt.xlabel("t-SNE1")
plt.ylabel("t-SNE2")

Text(0, 0.5, 't-SNE2')

inertias = []

for k in range(1, 30):
    # Building and fitting each model
    kmeans = KMeans(n_clusters=k, random_state=42)
    kmeans.fit(features_scaled)
    inertias.append(kmeans.inertia_)

# Plotting
plt.plot(range(1,30), inertias, 'bo-', label='Inertias')
plt.xlabel('Number of clusters (k)')
plt.ylabel('Distortion')
plt.title('Elbow Method')
plt.show()

from sklearn.metrics import silhouette_score

#calc score for ks 5 to 25 to help determine good k value
sil_scores = []
for k in range(5, 25):
    kmeans = KMeans(n_clusters=k, random_state=42)
    kmeans.fit(features_scaled)
    score = silhouette_score(features_scaled, kmeans.labels_)
    sil_scores.append(score)

plt.plot(range(5, 25), sil_scores, marker='o')
plt.title('Silhouette Score for Different k')
plt.xlabel('Number of Clusters')
plt.ylabel('Silhouette Score')
plt.show()

kmeans = KMeans(n_clusters=14, random_state=42)
kmeans.fit(features_scaled)
features["Cluster"] = kmeans.labels_
palette = sns.color_palette(cc.glasbey, n_colors=14)

tsne = TSNE(n_components=2, random_state=42)
features_tsne = tsne.fit_transform(features_scaled)

sns.scatterplot(x=features_tsne[:, 0],y=features_tsne[:, 1], hue="Cluster", data=features, s=45, palette=palette)
plt.legend(title="Centroid Groups", bbox_to_anchor=(1.05, 1), loc="upper left")
plt.title("Cluster Distribution")
plt.xlabel("t-SNE1")
plt.ylabel("t-SNE2")

Text(0, 0.5, 't-SNE2')

centroids = kmeans.cluster_centers_
centroids_df = pd.DataFrame(centroids, columns=features.columns)
centroids_df = centroids_df.drop(columns=["Cluster"])

#save the main feature for later
cluster_mapping = {}

#plotting feature contributions
for i, centroid in enumerate(centroids_df.values):
    plt.figure(figsize=(14, 6))
    sns.barplot(x=centroids_df.columns, y=centroid)

    cluster_mapping[i] = centroids_df.columns[np.argmax(centroid)]

    plt.title(f"Cluster {i} Features")
    plt.xlabel("Feature")
    plt.ylabel("Feature Value")
    plt.xticks(rotation=50)
    plt.tight_layout()
    plt.show()

cluster_mapping

{0: 'Pets & Animals',
 1: 'Howto & Style',
 2: 'Entertainment',
 3: 'Film & Animation',
 4: 'Education',
 5: 'Gaming',
 6: 'Comedy',
 7: 'People & Blogs',
 8: 'Music',
 9: 'Science & Technology',
 10: 'Autos & Vehicles',
 11: 'News & Politics',
 12: 'Travel & Events',
 13: 'Sports'}

#Plot rolling mean of cluster sizes over time
cluster_size_rolling = features.groupby(["year", "month__sin", "month_cos", "day_sin", "day_cos", "hour_sin", "hour_cos", "Cluster"]).size().unstack().rolling(window=30, min_periods=1).mean()

clusters = cluster_size_rolling.columns
years = cluster_size_rolling.index.get_level_values("year")
#Used so our x-axis doesn't have a tick for each point
x_axis_spacing = int(len(cluster_size_rolling) / len(years.unique()))

#Create a plot for each cluster
for cluster in clusters:
    plt.figure(figsize=(15, 8))
    cluster_size_rolling[cluster].plot(kind="line")
    plt.title(f"Rolling Average Cluster Size Over Time for Cluster {cluster}, Main Feature: {cluster_mapping[cluster]}")
    plt.xlabel("Time")
    plt.ylabel("Rolling Average Cluster Size")
    plt.xticks(ticks=range(0, len(cluster_size_rolling), x_axis_spacing), labels=years[::x_axis_spacing], rotation=45)
    plt.grid(True)
    plt.show()

	Title	Video URL	Channel	Date Watched	Ad?
0	Watched https://www.youtube.com/watch?v=ifoa0L...	https://www.youtube.com/watch?v=ifoa0LKZaXo	NaN	2024-10-21T23:36:50.230Z	True
1	Watched https://www.youtube.com/watch?v=F-Qc8f...	https://www.youtube.com/watch?v=F-Qc8fNrbzY	NaN	2024-10-21T23:36:45.823Z	True
2	Watched https://www.youtube.com/watch?v=1Lr8Zu...	https://www.youtube.com/watch?v=1Lr8ZuquO-Q	NaN	2024-10-21T23:36:38.805Z	True
3	Viewed Ads On YouTube Homepage	No URL	NaN	2024-10-21T23:36:25.100Z	True
4	Watched https://www.youtube.com/watch?v=c4xEw5...	https://www.youtube.com/watch?v=c4xEw53CYpQ	NaN	2024-10-21T23:36:21.557Z	True
5	Watched https://www.youtube.com/watch?v=hCXxAJ...	https://www.youtube.com/watch?v=hCXxAJrc-Os	NaN	2024-10-21T23:35:23.987Z	False
6	Watched DAVID ENG H277773 EQUITY EARBUDS US 20...	https://www.youtube.com/watch?v=i2vPSRmgqUo	NaN	2024-10-21T23:35:17.975Z	True
7	Watched https://www.youtube.com/watch?v=Ef8vb7...	https://www.youtube.com/watch?v=Ef8vb7mF8a4	NaN	2024-10-21T23:34:10.741Z	False
8	Watched https://www.youtube.com/watch?v=I8zKFw...	https://www.youtube.com/watch?v=I8zKFwzpjZU	NaN	2024-10-21T23:34:05.141Z	False
9	Watched Let's Rank the Leviathans in Subnautic...	https://www.youtube.com/watch?v=BPI0MhG-aqw	Floydson	2024-10-21T23:33:06.309Z	False

	Title	Video URL	Channel	Date Watched	Ad?
0	https://www.youtube.com/watch?v=hCXxAJrc-Os	https://www.youtube.com/watch?v=hCXxAJrc-Os	NaN	2024-10-21T23:35:23.987Z	False
1	https://www.youtube.com/watch?v=Ef8vb7mF8a4	https://www.youtube.com/watch?v=Ef8vb7mF8a4	NaN	2024-10-21T23:34:10.741Z	False
2	https://www.youtube.com/watch?v=I8zKFwzpjZU	https://www.youtube.com/watch?v=I8zKFwzpjZU	NaN	2024-10-21T23:34:05.141Z	False
3	Let's Rank the Leviathans in Subnautica: Below...	https://www.youtube.com/watch?v=BPI0MhG-aqw	Floydson	2024-10-21T23:33:06.309Z	False
4	https://www.youtube.com/watch?v=tREaghvgqiw	https://www.youtube.com/watch?v=tREaghvgqiw	NaN	2024-10-21T23:28:53.581Z	False
...	...	...	...	...	...
30833	https://www.youtube.com/watch?v=pkjGVzhGEN4	https://www.youtube.com/watch?v=pkjGVzhGEN4	NaN	2018-03-27T00:46:59.455Z	False
30834	SPIDER-MAN ROCK CLIMBING DISASTER (Cell Outs)	https://www.youtube.com/watch?v=IyAos6ddX5A	Smosh Games	2018-03-27T00:38:58.436Z	False
30835	LOGAN REAL LIFE TRAINING (Cell Outs)	https://www.youtube.com/watch?v=1nBtDv00GgQ	Smosh Games	2018-03-27T00:25:51.738Z	False
30836	ARCHERY PAINTBALL TAG (Cell Outs)	https://www.youtube.com/watch?v=A7g23cQlWtc	Smosh Games	2018-03-27T00:18:44.381Z	False
30837	TOMAHAWK CHOP IS OUR DEATHBLOW (Cell Outs)	https://www.youtube.com/watch?v=gfVxlblJReg	Smosh Games	2018-03-26T22:54:34.867Z	False

	Title	Video URL	Channel	Category	View Count	Like Count	Comment Count	Duration
0	Magnus Carlsen: "I'm No Genius’" 😮	https://www.youtube.com/watch?v=hCXxAJrc-Os	GJ_Chess	People & Blogs	1247761.0	57619.0	1469.0	PT51S
1	The MOST VALUABLE ITEM in Stardew Valley 1.6?! 😱	https://www.youtube.com/watch?v=Ef8vb7mF8a4	MRIZ	Gaming	46724.0	3514.0	28.0	PT1M
2	Michael Wants To Sleep with Poki and Lily??	https://www.youtube.com/watch?v=I8zKFwzpjZU	Offline Network	Entertainment	20657.0	2211.0	15.0	PT54S
3	Let's Rank the Leviathans in Subnautica: Below...	https://www.youtube.com/watch?v=BPI0MhG-aqw	Floydson	Gaming	2049923.0	126350.0	752.0	PT56S
4	My last two brain cells trying to do this chal...	https://www.youtube.com/watch?v=tREaghvgqiw	OfflineTV	News & Politics	111920.0	6318.0	46.0	PT34S
...	...	...	...	...	...	...	...	...
28414	TRAPEZE CIRCUS FAIL (Cell Outs)	https://www.youtube.com/watch?v=pkjGVzhGEN4	Smosh Games	Gaming	493012.0	9738.0	775.0	PT19M44S
28415	SPIDER-MAN ROCK CLIMBING DISASTER (Cell Outs)	https://www.youtube.com/watch?v=IyAos6ddX5A	Smosh Games	Gaming	297159.0	6968.0	861.0	PT11M17S
28416	LOGAN REAL LIFE TRAINING (Cell Outs)	https://www.youtube.com/watch?v=1nBtDv00GgQ	Smosh Games	Gaming	460051.0	9020.0	616.0	PT24M26S
28417	ARCHERY PAINTBALL TAG (Cell Outs)	https://www.youtube.com/watch?v=A7g23cQlWtc	Smosh Games	Gaming	390391.0	8607.0	484.0	PT13M27S
28418	TOMAHAWK CHOP IS OUR DEATHBLOW (Cell Outs)	https://www.youtube.com/watch?v=gfVxlblJReg	Smosh Games	Gaming	517716.0	11595.0	1417.0	PT17M5S

	Title	Video URL	Channel	Date Watched	Ad?
486	Answered survey question	No URL	Question: Uyavuma yini ukuthi isikhangiso sang...	2024-08-09T18:42:32.939Z	False
1214	Answered survey question	No URL	Question: Sicela ulinganisele okwenzeke kuwe n...	2024-04-07T15:13:34.687Z	False
1480	Answered survey question	No URL	Question: 過去一週內，您是否曾研究濾水器或淨水器的購買資訊？	2024-01-25T15:40:08.936Z	False
2354	Answered survey question	No URL	Question: Uyavuma yini ukuthi isikhangiso sang...	2023-10-28T03:51:41.926Z	False
2461	Answered survey question	No URL	Question: Uyavuma yini ukuthi isikhangiso sang...	2023-09-22T01:40:46.129Z	False
3044	Answered survey question	No URL	Question: Uyavuma yini ukuthi isikhangiso sang...	2023-06-08T16:18:24.704Z	False
3140	Answered survey question	No URL	Question: Uyavuma yini ukuthi isikhangiso sang...	2023-04-12T23:04:44.778Z	False
3307	Answered survey question	No URL	Question: Ingabe ziphinde kangakanani izikhang...	2023-01-08T02:41:44.320Z	False
3343	Answered survey question	No URL	Question: Ingabe uzizwa kanjani ngokuphindaphi...	2022-12-16T23:31:22.290Z	False
3441	Answered survey question	No URL	Question: Ingabe uyavuma ukuthi inani lezikhan...	2022-11-15T00:24:21.305Z	False
6067	Answered survey question	No URL	Question: How do you feel about the number of ...	2021-04-10T20:14:25.268Z	False
6359	Answered survey question	No URL	Question: 年齢を教えてください（ご自身や他のユーザーに表示される動画や広告の関連性...	2021-02-23T20:54:44.177Z	False
6582	Answered survey question	No URL	Question: 年齢を教えてください（ご自身や他のユーザーに表示される動画や広告の関連性...	2021-01-17T23:52:48.597Z	False
7235	Answered survey question	No URL	Question: 年齢を教えてください（ご自身や他のユーザーに表示される動画や広告の関連性...	2020-11-18T20:53:43.925Z	False
9236	Answered survey question	No URL	Question: 年齢を教えてください（ご自身や他のユーザーに表示される動画や広告の関連性...	2020-08-17T22:32:10.336Z	False
13273	Answered survey question	No URL	Question: 年齢を教えてください（ご自身や他のユーザーに表示される動画や広告の関連性...	2020-04-20T21:34:33.686Z	False
14932	Answered survey question	No URL	Question: 年齢を教えてください（ご自身や他のユーザーに表示される動画や広告の関連性...	2020-02-24T03:41:21.530Z	False
15998	Answered survey question	No URL	Question: 年齢を教えてください（ご自身や他のユーザーに表示される動画や広告の関連性...	2019-12-21T20:10:18.522Z	False

	Title	Video URL	Channel	Date Watched	Ad?
145	https://www.youtube.com/watch?v=BFG3WUW5QV8	https://www.youtube.com/watch?v=BFG3WUW5QV8	NaN	2024-09-22T18:53:23.950Z	False
185	https://www.youtube.com/watch?v=T7Jlm_P4HwU	https://www.youtube.com/watch?v=T7Jlm_P4HwU	NaN	2024-09-19T20:12:53.188Z	False
740	https://www.youtube.com/watch?v=TQSi1cuEeoM	https://www.youtube.com/watch?v=TQSi1cuEeoM	NaN	2024-06-30T00:18:10.804Z	False
976	https://www.youtube.com/watch?v=9pHT46WrXKQ	https://www.youtube.com/watch?v=9pHT46WrXKQ	NaN	2024-05-20T01:53:34.764Z	False
1000	https://www.youtube.com/watch?v=dxLvfHS4j_4	https://www.youtube.com/watch?v=dxLvfHS4j_4	NaN	2024-05-17T06:59:20.835Z	False
...	...	...	...	...	...
30733	https://www.youtube.com/watch?v=UTsbmlf-rv8	https://www.youtube.com/watch?v=UTsbmlf-rv8	NaN	2018-04-04T22:04:30.469Z	False
30739	https://www.youtube.com/watch?v=7mHa4q7RlaE	https://www.youtube.com/watch?v=7mHa4q7RlaE	NaN	2018-04-02T15:39:24.734Z	False
30755	https://www.youtube.com/watch?v=femunQQTrk8	https://www.youtube.com/watch?v=femunQQTrk8	NaN	2018-04-01T17:49:04.087Z	False
30763	https://www.youtube.com/watch?v=con9JUhk7s0	https://www.youtube.com/watch?v=con9JUhk7s0	NaN	2018-04-01T15:54:22.795Z	False
30767	https://www.youtube.com/watch?v=T00aqiuTnUg	https://www.youtube.com/watch?v=T00aqiuTnUg	NaN	2018-03-30T00:03:11.960Z	False

What does your watch history tell you?¶

Introduction¶

Data Curation¶

Exploratory Analysis¶

Further Analysis and Model Implementation¶

Insights & Conclusions¶

	View Count	Like Count	Comment Count	Duration (Min)
count	2.839300e+04	2.837300e+04	2.777800e+04	28419.000000
mean	6.649067e+06	1.487695e+05	7.721768e+03	18.572040
std	3.708439e+07	5.059813e+05	3.300550e+04	37.782237
min	7.000000e+00	0.000000e+00	0.000000e+00	0.000000
25%	3.570960e+05	9.880000e+03	5.820000e+02	7.133333
50%	1.405756e+06	3.602700e+04	1.831000e+03	12.983333
75%	4.709618e+06	1.254570e+05	5.895750e+03	20.216667
max	3.730024e+09	1.833026e+07	2.368414e+06	1420.016667

	View Count	Like Count	Comment Count
Category
Autos & Vehicles	6.951393e+05	11070.800000	860.700000
Comedy	7.037875e+06	183132.138760	10255.543056
Education	3.287391e+06	116505.432778	6606.301158
Entertainment	9.278229e+06	200567.937864	9308.312639
Film & Animation	8.714130e+06	210747.325751	13541.422078
Gaming	3.338422e+06	114072.845489	5522.224938
Howto & Style	6.186834e+06	131191.079585	7753.477056
Music	3.616509e+07	463825.213873	25565.373541
News & Politics	7.288559e+06	162131.517647	8166.734940
Nonprofits & Activism	1.035038e+07	184121.666667	10098.200000
People & Blogs	3.596483e+06	91132.625708	4740.905779
Pets & Animals	3.183866e+06	62166.791014	4639.040378
Science & Technology	1.294245e+07	285639.552632	11599.234513
Sports	4.104786e+06	75600.013889	4370.030641
Trailers	1.009833e+06	526.333333	33.000000
Travel & Events	3.769180e+06	48717.917910	2804.491045